5 ms·
(1) People don't genuinely want to be evaluated on their predictions. If they did, they would (a) make probabilistic predictions and (b) make well-defined predi
by bo1024 4y ago
(1) People don't genuinely want to be evaluated on their predictions. If they did, they would (a) make probabilistic predictions and (b) make well-defined predictions. Example "with 80% probability, there will be more than one million Americans over 100 years old in 2040."
(2) Prediction on long timelines is too hard. It's more useful to generate ideas about possibilities. With the small samples here, people are almost just as likely to be lucky as right.
We saw a microcosm of this early in the pandemic, where some groups got really excited about forecasting the number of cases. (1) I didn't see anybody quantitatively evaluate their own predictions and own up to how good or bad they were. (2) Everyone pretty quickly gave up even trying to make predictions over a month away. It was too hard. So it devolved into fitting short-term curves and then I think mostly they gave up forecasting altogether.
- seizethecheese 4y agoTrue for most people. The exception to this rule is Bryan Caplan who is covered in this article.
- bombcar 4y agoAnd if you make a well-defined prediction, you should be able to work backwards from it to smaller predictions over intervening years. If, say, I were to predict that 40% of intercity travel would be via high speed rail in 2100 in California, I should be able to either argue why high speed rail would spring out of nowhere in 2099 or have a plausible story as to how it might grow.
- goatlover 4y agoAt least Vernor Vinge gave a date range for the singularity. Initially, he said he would be surprised if it happened before 2000 and after 2020. I think he's moved that goalpost forward a decade or two since. Kurzweil just says 2045, because that's where his chart or accelerating returns tells him it will happen.
- bombcar 4y agoIt is always wise to make your date after your estimated death but before your customers.
- bee_rider 4y agoIs there supposed to be an implication of "if it doesn't happen by 2020 I don't think it will ever happen" in the Vinge prediction?
- abecedarius 4y agoThat didn't quite match my memory of Vinge, so fwiw I just checked the two earliest publications I know of: > Such trend-curves... extrapolated 30 or 40 years they are so high and steep that even the most naive futurist discounts their accuracy. (That's the closest therein to a specific prediction date, from 1983: https://imgur.com/a/HrUxGts https://imgur.com/a/HrUxGts) > let me more specific: I'll be surprised if this event occurs before 2005 or after 2030. From 1993: https://edoras.sdsu.edu/~vinge/misc/singularity.html https://edoras.sdsu.edu/~vinge/misc/singularity.html
- goatlover 4y agoOkay, maybe I remembered it wrong, or someone else gave that range. Vinge has about 8 more years to be surprised.
- TigeriusKirk 4y agoSpecifically he would be surprised if there isn't a computer with greater than human intelligence by 2030. It's interesting that 29 years after he said that, and 8 years remaining on the timeline, it's not completely out of the realm of possibility.
- abecedarius 4y agoOne of the example possibilities mentioned in one of his essays or interviews was, what if with some poorly-understood tweaks to neural network training, it started working much better at scale? At the time I thought he was really reaching with that one.